The sharing of video content on websites has developed into a worldwide phenomenon, supported by hundreds of thousands of websites. Video hosting sites, such as YOUTUBE™, currently have millions of users and tens of millions of videos. A user of a video hosting site may be interested in his/her friends' activities (e.g., viewing a video, uploading a video, etc.) so that he/she can determine videos to view based at least in part on his/her friends' activities. Existing systems that recommend videos to a user based off the activities of friends have numerous problems.
A first problem present in existing solutions is that they do not present the activities of the friends in an organized and useful way. For example, if the user has 200 friends and each friend views 5 videos in a day (for a total of 1,000 videos viewed by the user's friends), it is difficult to present the 1,000 activities to the user in a meaningful way. This problem is further frustrated by the limited display space of the video hosting site.
A second problem present in existing solutions is that they only rely on friends' activities on a single video hosting site when recommending a video to a user. If a user's friends are viewing videos on multiple websites, then the user is missing out on viewing many of the same videos as the user's friends.